Abstract:Distributed hypothalamic-midbrain neural circuits orchestrate complex behavioral responses during social interactions. How population-averaged neural activity measured by multi-fiber photometry (MFP) for calcium fluorescence signals correlates with social behaviors is a fundamental question. We propose a state-space analysis framework to characterize mouse MFP data based on dynamic latent variable models, which include continuous-state linear dynamical system (LDS) and discrete-state hidden semi-Markov model (HSMM). We validate these models on extensive MFP recordings during aggressive and mating behaviors in male-male and male-female interactions, respectively. Our results show that these models are capable of capturing both temporal behavioral structure and associated neural states. Overall, these analysis approaches provide an unbiased strategy to examine neural dynamics underlying social behaviors and reveals mechanistic insights into the relevant networks.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: how to identify the connection between neural activity and social behavior through the calcium fluorescence signals in the mouse social behavior network recorded by multi - fiber photometry (MFP). Specifically, the researchers proposed a state - space analysis framework, using dynamic latent variable models (including the continuous - state linear dynamical system (LDS) and the discrete - state hidden semi - Markov model (HSMM)) to characterize the mouse MFP data. These models aim to capture the behavioral time structure in social interactions (such as aggressive and mating behaviors) and the related neural states, thus providing an unbiased method to study the neural dynamics behind social behaviors and revealing mechanistic insights into the relevant networks.
### Research Background
Social interactions, such as aggression, defense, and mating, involve the coordinated activities of multiple distributed regions in the brain, among which the hypothalamus - midbrain neural circuit plays a particularly important role. Although in the natural environment, the brain uses multi - dimensional signals to coordinate discrete actions, a comprehensive understanding of the neural coding behind complex, natural social behaviors remains an open goal in systems neuroscience. The multi - fiber photometry technique (MFP), as an in - vivo calcium imaging method, can detect the average fluorescence intensity changes based on genetically - encoded calcium (Ca²⁺) indicators and multimode fibers, and is used to measure the synchronous firing population activities in multiple brain regions of freely - moving animals. However, MFP lacks cell - resolution, which poses a challenge in the face of complex natural behaviors.
### Research Methods
To overcome the above challenges, the researchers proposed an analytical method combining LDS and HSMM to infer the underlying neural dynamics from MFP recordings. The specific steps are as follows:
1. **Experimental Design**: Use a custom - made high - density multi - fiber array to record the Ca²⁺ activities in 13 brain regions expressing estrogen receptor type 1 (Esr1+), which belong to a well - defined social behavior network (SBN).
2. **Data Analysis**:
- **LDS Method**: Directly operate on the Z - score - normalized ΔF/F activities, and apply the established preferential subspace system identification (PSID) inference method to identify the behavior - related and - unrelated latent states.
- **HSMM Method**: Based on the firing proxies inferred from Ca²⁺ activities, further apply a rank - invariant resampling step to convert them into Poisson - distributed count data, and the results are compared with the human - labeled behavior labels.
### Research Results
- **LDS Method**: It can predict neural observations with high precision, but has a weak predictive ability for rapid dynamics or sudden changes in behavior tracking.
- **HSMM Method**: It can not only capture the time structure and the hierarchical structure of aggressive behaviors, but also discover micro - behavior states that are sometimes overlooked by human observers.
### Conclusion
Through these two latent variable models, the researchers have successfully revealed the connection between neural activity and behavior in the social behavior network, especially in aggressive and mating behaviors. These findings provide a new perspective for understanding the neural basis of social behaviors.